CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   oscillation of residues ! (http://www.cfd-online.com/Forums/fluent/121315-oscillation-residues.html)

rayan24 July 25, 2013 14:33

oscillation of residues !
 
I don't understand why the residues begin to oscillate from 6000 iterations in my case! before 6000 iterations the residues appear to decrease normally, but after starting to oscillate, is anyone has explanation please

flotus1 July 26, 2013 08:50

As always, more information is needed to provide a conclusive answer.

Usually, oscillating residuals indicate that the round-off accuracy of the machine is reached. At this point, the solution is really converged.

rayan24 July 26, 2013 09:03

Thanks for your replay

Here's the link of the residues

https://docs.google.com/file/d/0B3mx...it?usp=sharing

Thanks for help

Alimohamadi_nasr July 26, 2013 09:12

i think it can have two origin, the first one is that the number of your meshes is not enough and the second one is that you cannot solve it as a steady state and you should change it to transient mode. do you have reverse flow in your domain? if yes, what its percentage of reverse flow ?( the number of cell you have reverse flow/total number of cell on your boundary that you have reverse flow)

rayan24 July 26, 2013 09:19

I don't think it could come from the mesh because I have more than 3 million cells, and I think if it comes from the mesh I had to have these oscillations at the beginning of the simulation and not after 4000 iterations? no?!

No, I don't have a reversed flow

Alimohamadi_nasr July 26, 2013 09:36

I suggest change it to transient mode. i have mentioned that it can have two reasons.

rayan24 July 26, 2013 09:46

Thanks for your replay but I specify that similar simulations have been performed with steady mode :)

Alimohamadi_nasr July 26, 2013 09:50

can you determine difference between your two simulations? both of them have same boundary condition and same regime flow?

rayan24 July 26, 2013 09:56

I said that similar simulations are realised in the passe and not the same case! have been made ​​by other persones in the litiratures !

macfly July 26, 2013 22:27

Quote:

Originally Posted by Alimohamadi_nasr (Post 442172)
i think it can have two origin, the first one is that the number of your meshes is not enough and the second one is that you cannot solve it as a steady state and you should change it to transient mode. do you have reverse flow in your domain? if yes, what its percentage of reverse flow ?( the number of cell you have reverse flow/total number of cell on your boundary that you have reverse flow)

Hi Ali,

What do you have in mind when asking this question? Are you relating reverse flow to too coarse mesh or to steady/transient?

Alimohamadi_nasr July 26, 2013 22:43

Hello Macfly

no, about reverse flow, I have been only curious that it happens or not.

Far July 27, 2013 10:51

Quote:

Originally Posted by rayan24 (Post 442167)
Thanks for your replay

Here's the link of the residues

https://docs.google.com/file/d/0B3mx...it?usp=sharing

Thanks for help

Solution is converged. Did you check any other parameter of interest ?

rayan24 July 27, 2013 14:18

Thank you Sijal Ahmed for your replay,

if the solution is converged why Residues oscillate? I know, you can correct me if I'm wrong, that the convergence is achieved when Residues stabilize (ie no change or a very small change of residues), Can you confirme me this analysis of convergence please??

Occasionally when I check the result I perceive it is not converged (because i fixed 1e-3 not enough I suppose).

Far July 27, 2013 14:40

If you are not satisfied then it can be a problem of mesh.

What does convergence means? It is possible that in few cells you have problem in convergence due to mesh or some other flow problem and that may not be important for your simulation. So it will give you oscillatory convergence pattern. but I see there is not much oscillation from the mean value.

Moreover when we check convergence, it means

rayan24 July 27, 2013 15:05

Thanks for likk,
in fact the first explanation in the link is that the residues must be no changes :) .. apparently my mesh is constricted as well, but I'll try to redo the mesh then see

Far July 27, 2013 15:16

I would not suggest redoing mesh without any reason :)

rayan24 July 27, 2013 15:35

OK,

So could you confirm to me that my residues in top don't pose a problem and we can consider that the solution is converged in your opinion??

N.B : I'm begining in CFD modeling
Thanks

Far July 27, 2013 15:53

I think it is. What are you modelling? What parameters you are looking for?

rayan24 July 27, 2013 17:08

I model the aeraulic flow and temperature distribution in a vehicle...

is the nature of the residues in top are not embarrassing?? Have you met during your career in CFD similar convergence?

Far July 27, 2013 17:17

Yes many times while simulating in Fluent. In CFX either you will get fully convergence or no convergence at all, while Fluent tries, tries and sometimes for ever.

Maybe you should play with some parameters of solver like relaxation for high order terms, applying pseudo transients, hybrid initialization...


All times are GMT -4. The time now is 03:59.